Over the past 15 years, we have focused our efforts primarily on Caribbean Hispanics from the Dominican Republic because the frequency of late-onset Alzheimer's disease (LOAD) is higher than in non-Hispanic whites and blacks, the families are large, there are relatively few founders, and there is strong evidence of inbreeding. In the previous funding cycle we met our recruitment goals, identified functional mutations in SORL1, initiated a large-scale genome wide association study (GWAS) and begun targeted re-sequencing and whole exome sequencing in Caribbean Hispanic families. Genetic analysis of LOAD has focused on common genetic variants using GWAS in unrelated patients and controls that may explain ~35% of the genetic variance of LOAD. Studies of rare variants are now beginning to emerge, particularly in family-based cohorts. As a result of preliminary work over the three previous funding cycles, we are in a very strong position for this renewal: 1) We have assembled a large cohort of unrelated Caribbean Hispanic LOAD cases and elderly controls and have identified several hundred families multiply affected by LOAD. 2) We investigated genome- wide (global burden) analyses of runs of homozygosity showing that the average run is significantly longer in cases than controls and stronger yet in familial LOAD, suggesting the existence of novel recessive LOAD gene(s). A locus-specific ROH analysis also revealed a genome-wide significant consensus region on chr9q34. 3) We have initiated a large scale GWAS of all families and a case-control cohort that will include over seven thousand subjects. A preliminary analysis has identified several putative associations with LOAD. 4) We have begun an analysis of data from a preliminary targeted re-sequencing of LOAD candidate genes, and whole exome sequencing (WES) in 115 members of 31 multiplex families. 5) Another 353 members of 67 multiplex families from this same cohort were included in the whole genome sequencing (WGS) study of the NIA/NHGRI Alzheimer's Disease Sequencing Project and analyses of these data are underway. Results of both analyses will be available to us near or just after the start date of this revised proposal. To take advantage of these resources and capitalize on our preliminary findings, we propose to use the results from the GWAS (Hispanic and non-Hispanic), analyses of runs of homozygosity, WES and WGS analyses from our population of Caribbean Hispanics, to conduct targeted re-sequencing in genes (loci), or pathways showing putative relationships to LOAD, followed by replication and validation genotyping, in silico bioinformatic and biophysical analyses of variant sequences identifying effects on gene structure and biological assays of gene function. The central hypothesis we are testing is that, as yet unidentified, common and rare genetic variants underlie the pathobiology of LOAD and can be found in large well-phenotyped multiplex families, cases and controls.